Implementasi Algoritma YOLO (You Only Look Once) Untuk Deteksi Rias Adat Nusantara

Abstract

Listen

Translate article

Makeup and fashion are important things that need to be considered by the future bride and groom. Traditional bridal makeup and dress in Indonesia have certain meanings and symbols according to the region and beliefs of their ancestors. In general, the meaning of the symbol is prayers and hopes that the home life to be lived will always be endowed with happiness and well-being. One way to distinguish the types of traditional makeup can be observed through the form of makeup on the face, the order and accessories used on the bride's hair. Current technological developments can be used for the introduction of customary types of makeup, namely through object detection by applying machine learning algorithms, one of which is YOLO (You Only Look Once). In this study, we will use yolo version 4 Tiny (YOLOv4-Tiny) as an algorithm to detect the type of nusantara customary makeup on the image of the bride's head by conducting a dataset training process on 1478 images or images consisting of 14 types of Nusantara customary makeup in 5000 iterations, the application built was able to obtain a value with an average accuracy of 95.20% and an average time needed to detect for 327ms (milli-seconds).